FRAUD DETECTION AT PT NESINAK INDUSTRIES IS SEEN FROM TIME PRESSURE AND WORK EXPERIENCE

This study aims to obtain empirical evidence on the effect of work experience, time pressure on fraud detection. For this study using quantitative type of research. This research was conducted using questionnaires distributed to employees working at PT Nesinak, as many as 116 people filled out questionnaires that were distributed directly. To get the results of this study, researchers used PLS SEM version 3.0. The results of this study found that work experience had a positive and statistically significant effect on fraud detection, but time pressure had a negative but statistically not significant effect on fraud detection. This research focuses on fraud detection at PT Nesinak that involves all parts.


INTRODUCTION
Employees or human capital become very important because they can generate added value for the company.The good name of the company is also influenced by the performance and behavior of employees.If employees have poor performance and behavior, it will have a bad impact on the company's image.For example, if there are employees who violate company regulations or commit fraud, it will have a fatal impact on the company's image both in the eyes of the public and company colleagues (Saputra, 2022).
The increasing prevalence of fraud that occurs in various ways that continue to grow requires companies to continue to improve the ability of their employees to detect fraud.The number of fraud cases that occur has attracted a lot of media attention as a prominent and important issue in the eyes of businesses and a highlight for all circles in society.Fraud can be done if employees have the opportunity and power.Employees who have high power are more likely to commit fraud (R. C. N. Harahap, 2020).
Fraud is one of the fraud acts that is often found and difficult to overcome.Fraud in a company is usually caused by a weak company control system.According to (R. C. N. Harahap, 2020), an employee commits fraud due to several influencing factors, namely pressure, opportunity and self-justification for the fraud committed.That way, it is important for a company to detect to prevent fraud (Faisal and Sari 2020).
In Bekasi itself, in 2019 there was a case of fraud committed by an employee at a wellknown retail company in Indonesia, namely PT.Source: Alfaria Trijaya.A cashier manipulates the sales transactions of store customers who are helped by relatives.This happened for 7 months starting from April 2018 -November 2018.This case was uncovered after the company audited the sales results and it was found that there was a discrepancy between the outgoing goods and sales and as a result of this incident, Alfamart was estimated to have suffered a loss of Rp.177,000,000.we can assume that this case occurred because of the opportunity that the cashier had so that his jobdesk was able to collude fraudulent acts.(www.metro.tempo.com)Another fraud also occurred in the Bekasi city government, namely fraud committed by the inactive Bekasi mayor, Rahmat Effendi, in early 2022 for committing fraud in the form of bribes for goods and services procurement projects and position auctions.Rahmat Effendi received Rp 7.1 billion from officials to state civil servants (ASN) within the Bekasi city government with the mode of debt repayment.This case can occur because the perpetrator has very high power and uses a very large opportunity to commit fraud.The impact of the incident was that people thought that there were weaknesses in identifying fraud, causing losses and loss of public trust in Bekasi city government agencies.(www.dakta.com).To improve the ability to detect fraud, you must pay attention to several things, namely time pressure and work experience.The greater the time pressure faced, the more influential it is on a person's ability to detect fraud.The more work experience, the better the ability to detect fraud (Elfia &;NR, 2022).
Another fraud also occurred in the Bekasi city government, namely fraud committed by the inactive Bekasi mayor, Rahmat Effendi, in early 2022 for committing fraud in the form of bribes for goods and services procurement projects and position auctions.Rahmat Effendi received Rp 7.1 billion from officials to state civil servants (ASN) within the Bekasi city government with the mode of debt repayment.This case can occur because the perpetrator has very high power and uses a very large opportunity to commit fraud.The impact of the incident was that people thought that there were weaknesses in identifying fraud, causing losses and loss of public trust in Bekasi city government agencies.(www.dakta.com).To improve the ability to detect fraud, you must pay attention to several things, namely time pressure and work experience.The greater the time pressure faced, the more influential it is on a person's ability to detect fraud.The more work experience, the better the ability to detect fraud (Elfia &;NR, 2022).
Employees who have a lot of working hours will be more capable and fast in detecting fraud or errors.This ability cannot be obtained instantly.Expertise can be obtained the more often we do a task.Someone who often does a task, then has experience to detect problems or mistakes that will occur, so that fraud can be identified (Irawan et al., 2018).In research conducted by (Faradina, 2016;R. J. Harahap &;Tobing, 2020;Indrayani et al., 2019;Irawan et al., 2018;Kusumawaty &;Betri, 2019;Masri et al., 2022;Novita, 2015;Rafnes &;Primasari, 2020;Tandijono et al., 2018;Yuanita &;Amanah, 2018) who examined the effect of work experience on the ability to detect fraud, showed that work experience has a positive effect on the ability to detect fraud.
Based on the explanation above, there are different results shown by previous researchers.Differences in research results can occur for several reasons such as differences in research time, researchers' interpretation of cheating, differences in variables used and differences in the use of testing methods.This research is interesting to do because there are many differences in results regarding this study.Based on this background, researchers are interested in conducting research entitled the effect of time pressure and work experience on fraud detection The Effect of Time Pressure on Fraud Detection The time pressure given forces employees to complete tasks as soon as possible or in accordance with the deadlines given, this can be pressure for employees because if someone completes their work beyond the specified time limit then they tend to be judged to have poor performance by their superiors.In order to hit deadlines there is a possibility for employees to make mistakes, mistakes and violations.(Anggriawan, 2014;Dandi et al., 2017;Elfia & NR, 2022;Pangestika et al., 2014;Salsabil, 2019;Sofie &;Nugroho, 2019;Yuanita &;Amanah, 2018) have proven in their research that time pressure negatively affects fraud detection because time pressure makes employees have to adjust time to the tasks to be completed.If the estimated time is not in accordance with what is actually needed, it tends to be less careful and prioritize some tasks in order to complete the task according to the deadline.However, the research conducted (Francisco et al., 2021;Laitupa &;Hehanussa, 2020;Moalina &;Wulandari, 2018) showed contradictory results where they concluded that time pressure has a positive effect on fraud detection.Based on this explanation, in this study the following hypotheses will be formulated; H 1 : Time pressure negatively affects fraud detection.

The Effect of Work Experience on Fraud Detection
In work, someone who is competent is required to have high expertise and professionalism, these skills can be obtained by the more often we do the task, namely experience.Someone who has high flying hours must already know the problems or mistakes that will occur, so mistakes and misappropriation will be avoided.

IMPLEMENTATION METHOD
The purpose of this study is to determine whether there is a significant influence between variables.The paradigm used in this study is positivism as a method that is arranged systematically using logic in finding conjectures about human cause and effect, because it usually only observes the visible surface without understanding it more deeply.The data used in this study used primary data, namely by giving questionnaires to respondents (Afriady &;Alfiansyah, 2022).This research method uses quantitative methods that have significant relationships between the variables studied to obtain conclusions that will explain the general picture under study in the form of values or scores for answers given to respondents to statements on questionnaires (Afriady &;Alfiansyah, 2022).The sampling design in this study is probability sampling.For the research background, researchers have no intervention in the research (noncontrived setting).For the implementation time using one (cross-section) using data analysis, namely hypothesis testing The subject of this study is an employee who works at PT Nesinak (Efendi 2022).
The qualifications of respondents who will be used in the researcher questionnaire are respondents of PT Nesinak employees.Based on variables (Hair et al., 2009) state that the amount of drinking of the sample is based on the summation of the highest indicators of each variable multiplied by five to ten (Faisal et al. 2021;Faisal and Sudibyo 2020).This study has 23 indicators in the study, so that the number of samples in this study multiplied by five (5) then the number of samples = 23 x 5 = 115 or at least 115 respondents.Respondents who filled out the questionnaire were employees of PT Nesinak.Furthermore, the questionnaires obtained and then sampled in the study were all questionnaires received again after being filled in by 116 respondents.

RESULTS AND DISCUSSION
The respondents who participated in this study were all employees at PT Nesinak.The data used is primary data, namely by distributing questionnaires to respondents who meet the criteria.The table above provides an explanation that this researcher has succeeded in collecting respondents in this study as many as 116 each respondent has answered all questions asked by researchers about the effect of time pressure, work experience and fraud prevention.Based on the results of data processing, respondent data was obtained as follows:

Figure 1 Respondent's Gender
Source: Researcher ( 2023) From the data obtained during the distribution of the questionnaire, it was obtained that the gender of respondents was male as many as 70 people and 46 people were female.So it can be seen that respondents who are male are more dominant than women at PT Nesinak.The study also found about the length of work of respondents.From the data collected, it was found that 78 people worked at PT Nesinak for more than 5 years.Which means that the majority of respondents have become permanent employees at PT Nesinak, only 13 people answered that they worked less than 1 year.As for respondents who answered 3-5 years, there are also quite a lot, namely there are 10 people and the remaining 15 people they work 1-2 years.

Figure 2 Length of Work of Respondents
Source : Researcher (2023) In addition to gender, the length of work of the respondents that the researchers found.The study also found the age of respondents working at PT Nesinak.Namely can be seen in the following figure table  From the data above, it was found that the age of respondents who worked at PT Nesinak as many as 53 respondents were over 35 years old.For the age of respondents 31 -35 years, there were 21 people.Only 3 people are less than 20 years old.And the remaining 20-25 years as many as 34 people.
The following tests are carried out on the outer model (measurement model), which can be known through the following stages:

Convergent Validity
Convergent validity, measurement with reflexive indicators is rated 96 based on the correlation between item score / component score with construct score.Individual reflexive measures are said to be high when correlated with more than 0.70.Convergent validity relates to the principle that measurements of a construct should be highly correlated.Convergent validity occurs when scores obtained from two different instruments that measure the same construct have a high correlation.The convergent validity test in PLS with reflective indicators is assessed fundamentally on the loading factor (correlation between component item scores and construct scores) indicators that measure the construct (Abdillah &;Hartono, 2015: 195).There are two criteria to assess whether the outer model meets the convergent validity requirements for reflective constructs, namely outer loading > 0.7, and average variance extracted (AVE) > 0.5, then the items in the variable are considered to have sufficient convergent validity (Hair et al., 2010).But in some cases, often the loading requirement above 0.7 is often not met, especially for newly developed questionnaires.Therefore, loading between 0.40-0.70 must still be considered and maintained (Mahfud and Ratmono, 2013: 66).Based on the results of data processing, convergent validity results are obtained with loading factor values, as follows:  Based on the results of the table above, convergent validity results are obtained with loading factors, all indicator loading factors have a value of 0.40-0.70,so they can be declared valid.On the other hand, convergent validity can also be measured by calculating each indicator on the average variance extracted (AVE).Indicator to calculate AVE, if the AVE value is more than 0.5 then the items in the variable are considered to have sufficient convergent validity.(Hair et al., 2010;Ghozali, 2008;Indrawati, 2015:153).The results of the AVE value can be seen in the table as follows: Based on the table above, the results of convergent validity calculations with AVE, obtain that the AVE value of each variable has a value of more than 0.50.So it can be stated that the data in this study have met the criteria of convergent validity.

Discriminant Validity
The validity of related discriminants occurs if two different instruments that measure two predicted constructs do not correlate resulting in an uncorrelated score (Hartono, 2008: 64).According to (Abdillah &;Hartono, 2015: 196), the discriminant validity test is assessed based on cross loading with measurements and constructs.Then the construct is declared to have discriminant validity.Discriminant validity, a value based on crossloading with the construct.If the correlation of the construct with the measurement item is greater compared to the size of other constructs, then it shows that the latent construct predicts the size on the block better than the size on the other block.Based on the results of data processing, dicriminant validity results with cross loading can be known through the table, as follows; In addition to the forms of cheating, I am also able to easily identify parties who have committed fraud.Based on the table above, it is obtained that each item has the highest correlation value of the other constructs.Therefore, the variables in this study can be declared to meet the criteria of discriminant validity.Another criterion used to test discriminant validity can be seen from the Root Square AVE value.if the square root value of each AVE variable is greater than the correlation between two 100 variables in the model, then the research questionnaire has a discriminant validity value.(Indrawati et al., 2017:70).The result of the value of the fornell -larcker criterion can be seen in the table, as follows: Based on the table above, discriminant validity with the Fornell Larcker criteria shows the AVE root value of each construct or variable, the AVE square root results of each variable are greater than the correlation between the two variables in the model.Therefore, the variables in this study can be declared to meet the criteria for discriminant validity.

Reliability Test
A reliability test must be carried out to find out whether each item on the questionnaire meets the reliability criteria.Reliability shows the accuracy, consistency and precision of a measuring instrument in carrying out a measurement (Hartono, 2008).According to (Abdillah & Hartono, 2015: 196) in conducting reliability tests in PLS there are two methods, namely Cronbach's alpha and Composite reliability.Cronbach's Alpha measures the lower limit of the reliability value of an item, while composite reliability measures the actual value of the reliability of a construct (Abdillah & Hartono, 2015: 196).The reliability test is strengthened by the expected Cronbach Alpha value.0.70 on each indicator.Based on the results of data processing, the reliability results obtained with Cronbach's alpha and composite reliability can be seen in the table as follows: Based on the table above, the results of the reliability test, show that all variables in this study have cronbach alpha and composite reliability values and each has a value of > 0.7, which means that it has met the reliability test criteria In this study using structural models in PLS evaluated using R2 for dependent constructs, coefficient path values or t-values of each path for significance tests between constructs in structural models (Abdillah &;Hartono, 2015: 197) (Abdillah and Hartono 2015).According to Ghozali and Latan (2015), the inner model is often also called the inner relation model which describes the relationship between latent variables based on substantive theory.The design of structural models of relationships between latent variables is based on problem formulations or research hypotheses.In this model evaluation, estimation can be done through several stages.The use of R-square for dependent constructs, Stone-Geisser Q-Square test for predictive relevance and t test as well as significance of structural path parameter coefficients.In this study, the second test carried out was inner model testing.Inner model testing has 3 types of evaluations carried out, namely R-square, Q-square and path coeficient, using the help of SmartPLS software.The inner model path diagram in this study can be seen in the figure below: R² is used to measure the degree of variation in changes in independent variables to dependent variables, and the value of the path coefficient shows the level of significance in hypothesis testing (Abdillah &;Hartono, 2015: 197).The results of R 2 > 0.67 for endogenous latent variables in structural models indicate the influence of exogenous variables on endogenous variables in the good category.If the result is 0.33-0.67 it is in the medium category and if the result is 0.19-0.33 then it is in the weak category.Based on the results of data processing, rsquare results are obtained, as follows: Based on the table above, the results of the rsquare (R 2 ) value for the fraud detection variable were obtained at 0.615, which can be stated to be in the medium category, it shows that 61.5% contribution is the influence of time pressure, and work experience.
Relevant Predictions (Q-Square) Stone-Geisser Q-square test for predictive relevance.Q-square (Q2) measures how well the observation values are produced by the model and its parameter estimation.Q-qsuare is used to test how well the values produced by the model and to determine the estimated parameters.The Qsquare value is above 0 indicating that the model has a predictive relevance value while the Qsquare value is below 0 indicating that the model lacks a predictive relevance value (Ghozali, 2014: 41).R-square PLS models can be evaluated by looking at the Q-square predictive relevance for variable models.The Q-square measures how well the observation values are produced by the model and also the estimation of its parameters.A Q-square value greater than 0 (zero) indicates that the model has a predictive relvance value, while a Q-square value less than 0 (zero) indicates that the model lacks predictive relevance.However, if the calculation results show a Q-square value of more than 0 (zero), then the model deserves to be said to have a relevant predictive value.The results of calculating the Q-Square value are as follows: Based on the results of these calculations, the Q-square results were obtained by 0.615 or 61.5%, so it can be stated that the magnitude of the diversity of this research data is 61.5%, while the remaining 38.5% is explained by other variables outside this study.

Hypothesis Testing
The hypothesis in this study can be known from calculating the model using PLS bootstrapping technique.From the results of the bootstrapping calculation, a statistical t value of each relationship or path will be obtained.The testing of this hypothesis is set with a significance level of 0.05.The hypothesis is acceptable if the t-statistic value > table.The calculation results for hypothesis testing in this study, using the direct influence of the independent variable on the dependent variable obtained as follows:  Braun (2000) in Koroy (2008: 29), illustrating one of the effects of time pressure on the performance of working people in fraud detection, Braun pointed out that where people who work under time pressure, some tasks will be prioritized over others.Braun tested his hypothesis that when time pressure is increased in a multi-tasking environment, lower task performance (i.e.sensitivity to cheating cues) will decrease while dominant task performance (documenting evidence) will remain unchanged.The results showed that people who work at PT Nesinak who are under time pressure will be less sensitive to cheating cues and therefore less likely to be able to detect cheating.Thus people who work at PT Nesinak may lose evidence that will affect the results of the audit.A person who works at PT Nesinak is required to be able to work under time pressure without compromising the quality of his work, but in reality not everyone is able to do this.Although working under time pressure a person who works at PT Nesinak must be able to detect fraud and work professionally and act according to applicable standards.

Work experience positively affects fraud detection
Experience is an important indicator of a person at work.Work experience has been seen as an important factor in predicting performance, so experience is included in the requirements in working at PT Nesinak, in the work of knowledge carried out during education but no less important is the experience gained.Working people who already have experience are believed to be able to detect fraud that occurs in the world of work.Because of his experience in dealing with both reasonable and unnatural cases, the more experience a person has working in analyzing and overcoming problems, the more adept the person is at solving them.A person who has worked for a long time will have good experience and flying hours.This will affect their performance at work because they are used to cheating.The experience of people who have worked also makes it easier for someone to know about errors and fraud that occur so that the results will be more valid and the job report can be trusted.It can also increase the credibility of a person who works so that it can be viewed by other companies.

CONCLUSION
Based on this study, it can be concluded that the time pressure experienced by employees at PT Nesinak has a negative and statistically insignificant effect on fraud detection.With the pressure of employee time in carrying out their duties will be unfocused, rushed so that the work given is completed in accordance with the time given, it will be difficult to detect if there is fraud in the work process.Work experience of employees at PT Nesinak has a positive and statistically significant effect on fraud detection at PT Nesinak.Employees who work for a long time (at least three years) in their field have good work experience so that it can be easy to detect fraud, because employees will easily find if there are irregularities / abnormalities in the work process.
Based on the results of research that has been conducted, researchers provide advice that the time pressure faced by employees causes employees at PT Nesinak will prioritize some tasks if the budgeted time is lacking, this will reduce employee attention to fraud detection.So that this does not happen, in budgeting time should be superiors / leaders at PT Nesinak considers the complexity of the tasks assigned to his subordinates and does not make time budgets that are too tight and too loose.In detecting fraud, the experience of a person who works is very important because with adequate experience employees are accustomed to facing something in a process and better understand the symptoms of occurrence in terms of fraud, for that the experience of an employee must continue to be improved, for example by attending trainings in their fields and also training that supports their work processes, besides that junior employees are more often involved in tasks that complex so that it will be more experienced so that it will be easy to detect if fraud occurs. :

Figure 3
Figure 3 Age of Respondents

I
try to complete my task according to the agreed completion time 0,891 Valid I find the execution or completion of certain work within the time limit difficult to fulfill 0,912 Valid I felt burdened by the time limit because it put a lot of pressure on me.0,801ValidThe existence of agreements in the processing time makes the quality of work decrease.to be experienced if they have carried out their duties for more than 3 (three) years.0,883ValidA junior to achieve his competence can learn from the experience of his senior.0,831ValidThe longer I work, the more I can avoid mistakes.0,802 Valid Experience has certainly increased with frequent tasks.0,832 Valid Experience can help me in knowing the cheating committed by my colleagues.0,838ValidDifficult tasks can affect the ability to work.0,862 Valid The difficulty of assignments in the past is very beneficial for me in facilitating future work.knowledge of the types of cheating, especially those that often occur at the time of assignment.0,858ValidIn addition to the forms of cheating, I am also able to easily identify parties who have committed fraud.
by the time limit because it put a lot of pressure on me. the faster the work is completed, the better the quality of the work.difficulty of assignments in the past is very beneficial for me in facilitating future work.0,766 0,788 0,874 I have sufficient knowledge of the types of cheating, especially those 0at the time of assignment.

FRAUD
Figure 1.Inner Model

Table 2
Loading Factors

Tabel 3 hasil uji hipotesis
Based on the table above, the conclusion of the results of hypothesis testing is obtained, by comparing the results of t-statistics with ttable (1,973) with a significance level of 0.05, the conclusion of hypothesis testing is obtained, as follows: 1.The results of testing the hypothesis of the effect of time pressure on fraud detection, obtained tstatistic results of 0.512, then compared with ttable values and 5% probability, because the results of t-statistics<ttable (0.512<0.1.973)and significance 0.610 > 0.05, then H0 is accepted and Ha is rejected, so it can be stated that Time Pressure has a negative and insignificant effect on Fraud Detection.2. The results of testing the hypothesis of the effect of Work Experience on fraud detection obtained t-statistical results of 12.091, then compared with table values and probability of 5%, because the results of t-statistics>ttable (12.091>1.973)and significance 0.000 < 0.05, then H0 is rejected and Ha is accepted, so it can be stated that Work experience has a positive and statistically significant effect on fraud detection Time pressure affects fraud detection Someone who works under time pressure will have a reduced level of accuracy compared to people who work without pressure.Research by